Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. An imaging device for imaging an object, the imaging device comprising: an image pickup device configured to output an image signal, the image signal comprising a plurality of frames; a video signal processor configured to perform processing for generating a video signal based on the image signal; an image recognition unit, implemented by a hardware circuit, configured to perform a plurality of pieces of recognition processing on the image signal; and a luminance image creation unit, implemented by a hardware circuit, configured to extract a luminance image from the image signal output from the image pickup device, wherein the processing of the video signal processor and the processing of the image recognition unit are executed in parallel to each other, and wherein a number of bits of a digital signal corresponding to the image signal is reduced when the image recognition unit detects a preset state.
This invention relates to an imaging device designed to capture and process images of an object while optimizing computational efficiency. The device includes an image pickup device that outputs an image signal composed of multiple frames. A video signal processor generates a video signal from this image signal, while an image recognition unit, implemented in hardware, performs multiple recognition tasks on the image signal. Additionally, a luminance image creation unit, also implemented in hardware, extracts a luminance image from the image signal. The video signal processing and image recognition tasks operate simultaneously to enhance performance. When the image recognition unit detects a predefined state, the device reduces the bit depth of the digital signal corresponding to the image signal, conserving processing resources without compromising essential functionality. This approach balances real-time processing demands with computational efficiency, particularly useful in applications requiring both high-speed video output and simultaneous image analysis, such as surveillance or autonomous systems. The hardware-based implementation ensures low-latency operation, while the adaptive bit reduction optimizes power and processing efficiency under specific conditions.
2. The imaging device according to claim 1 , wherein the image recognition unit performs the plurality of pieces of recognition processing using RAW data output from the image pickup device.
This invention relates to an imaging device with enhanced image recognition capabilities, particularly for processing RAW image data directly from an image pickup device. The device includes an image pickup unit that captures images and outputs RAW data, which is unprocessed sensor data containing high-fidelity information. A recognition unit performs multiple recognition processes on this RAW data, such as object detection, scene classification, or facial recognition, without requiring prior conversion to a compressed or processed image format. This approach preserves the original image quality and detail, improving recognition accuracy. The recognition unit may also adjust recognition parameters based on the RAW data characteristics, such as noise levels or dynamic range, to optimize performance. The device may further include a processing unit that applies corrections or enhancements to the RAW data before recognition, ensuring consistent input quality. By processing RAW data directly, the system avoids degradation from intermediate compression or processing steps, leading to more reliable and precise recognition results. This method is particularly useful in applications requiring high accuracy, such as medical imaging, surveillance, or industrial inspection.
3. The imaging device according to claim 1 , wherein the image recognition unit comprises a plurality of filters for extracting different types of features from the luminance image, and the plurality of filters comprise filters based on learning by using a neural network.
This invention relates to an imaging device with enhanced image recognition capabilities, particularly for extracting and analyzing features from luminance images. The device addresses the challenge of accurately identifying and processing diverse visual features in images, which is critical for applications like object detection, scene understanding, and automated image analysis. The imaging device includes an image recognition unit designed to process luminance images by applying multiple filters. These filters are specialized for extracting different types of features from the image data. The key innovation lies in the use of filters that are trained through a neural network, enabling the device to learn and adapt to various feature extraction tasks. This approach improves recognition accuracy and flexibility compared to traditional fixed-filter methods. The neural network-based filters are trained to recognize patterns, textures, edges, and other relevant features in the image. By leveraging machine learning, the device can dynamically adjust its feature extraction process, making it more robust in varying conditions. This technology is particularly useful in applications requiring real-time image analysis, such as surveillance systems, autonomous vehicles, and medical imaging, where precise and adaptable feature recognition is essential. The use of multiple filters ensures comprehensive feature extraction, enhancing the overall performance of the imaging device.
4. The imaging device according to claim 3 , wherein processing by the plurality of filters comprises convolution processing.
The invention relates to an imaging device designed to enhance image quality through advanced processing techniques. The device incorporates a series of filters that apply convolution operations to the captured image data. Convolution processing involves applying a set of mathematical kernels to the image, which modifies pixel values based on neighboring pixels to achieve effects such as blurring, sharpening, edge detection, or noise reduction. This method allows for real-time or post-processing adjustments to improve visual clarity, reduce artifacts, or extract specific features from the image. The convolution-based filtering system operates within the imaging pipeline, either as part of the sensor processing unit or as a dedicated hardware accelerator, ensuring efficient and scalable image enhancement. By leveraging convolution, the device achieves precise control over image characteristics while maintaining computational efficiency, making it suitable for applications in photography, medical imaging, surveillance, and machine vision. The approach differs from traditional pixel-level adjustments by using neighborhood-based transformations, which provide more sophisticated and adaptive image processing capabilities.
5. The imaging device according to claim 1 , wherein the image recognition unit performs the plurality of pieces of recognition processing using color information included in the video signal generated by the video signal processor.
The technology domain involves imaging devices equipped with image recognition capabilities, specifically focusing on processing video signals to extract and utilize color information for recognition tasks. The problem addressed is enhancing the accuracy and efficiency of image recognition by leveraging color data inherent in video signals, which can improve object detection, scene understanding, or other recognition-based applications. The imaging device includes a video signal processor that generates a video signal containing color information. An image recognition unit processes this video signal by performing multiple recognition tasks, such as identifying objects, classifying scenes, or detecting features, using the color data embedded within the signal. This approach ensures that recognition processing is not solely dependent on luminance or structural information but also incorporates chromatic details, potentially increasing robustness and precision in varied lighting or environmental conditions. By integrating color information into the recognition pipeline, the device aims to achieve more nuanced and reliable interpretations of visual data, which can be particularly beneficial in applications like surveillance, autonomous navigation, or industrial inspection where color cues are critical. The system does not rely on external color calibration or additional sensors, as the color information is inherently derived from the video signal itself.
6. The imaging device according to claim 1 , wherein the image recognition unit comprises a plurality of filters for extracting different types of features from the luminance image, and the plurality of filters comprise filters provided for a static filter and filters provided for a dynamic filter, the static filter configured to extract features from a luminance image of one frame without motion, the dynamic filter configured to extract features related to motions by comparing information between frames of the image signal.
This invention relates to an imaging device with an enhanced image recognition unit designed to improve feature extraction from luminance images. The device addresses the challenge of accurately identifying and processing both static and dynamic features in image data, which is critical for applications like object detection, tracking, and scene analysis. The image recognition unit includes multiple filters that extract different types of features from the luminance image. These filters are categorized into static filters and dynamic filters. Static filters are configured to extract features from a single frame of the luminance image, focusing on stationary elements within the scene. Dynamic filters, on the other hand, are designed to detect motion-related features by comparing information between consecutive frames of the image signal. This dual-filter approach allows the device to capture both static and dynamic characteristics of the scene, improving the accuracy and robustness of image recognition tasks. By separating the extraction of static and dynamic features, the imaging device can more effectively analyze complex scenes where objects may be moving or stationary. This enhances the device's ability to perform tasks such as object tracking, gesture recognition, and environmental monitoring. The use of multiple filters ensures that a wide range of features is captured, providing a comprehensive understanding of the image content.
7. The imaging device according to claim 6 , wherein the image recognition unit summarizes the features extracted by the filters for the static filter and the filters for the dynamic filter, and recognizes a content of the luminance image based on a result of summarization.
This invention relates to an imaging device with enhanced image recognition capabilities, particularly for analyzing luminance images to distinguish between static and dynamic features. The device includes an image recognition unit that processes a luminance image by applying a set of filters to extract features. These filters are categorized into static filters, which detect features in static regions of the image, and dynamic filters, which detect features in dynamic regions. The image recognition unit then summarizes the extracted features from both filter types and uses this summarized data to recognize the content of the luminance image. This approach improves accuracy by separately analyzing static and dynamic elements, allowing for better differentiation between background and moving objects. The device may also include a luminance image generation unit that converts a captured image into a luminance image, ensuring consistent input for feature extraction. The system is designed to enhance object recognition in environments with both static and dynamic elements, such as surveillance or autonomous navigation, by leveraging specialized filters and summarization techniques.
8. The imaging device according to claim 7 , whether the image recognition unit recognizes the content of the luminance image by determining an object corresponding to the features extracted by the filters for the static filter and determining that a motion mode of the object coincides with a motion mode corresponding to the features extracted by the filters for the dynamic filter.
This invention relates to an imaging device with enhanced image recognition capabilities, particularly for distinguishing static and dynamic features in captured images. The device addresses the challenge of accurately identifying objects and their motion states in real-time imaging applications, such as surveillance, autonomous navigation, or robotics, where distinguishing between stationary and moving objects is critical. The imaging device includes an image recognition unit that processes luminance images to extract features using both static and dynamic filters. The static filters identify features associated with stationary objects, while the dynamic filters extract features linked to motion patterns. The recognition unit then determines whether the content of the luminance image corresponds to an object by matching the extracted static features. Additionally, the device assesses whether the object's motion mode aligns with the motion features identified by the dynamic filters. This dual-filter approach improves recognition accuracy by correlating static and dynamic characteristics, ensuring reliable object detection and motion analysis. The system enhances performance in environments where objects may partially occlude or exhibit complex movement, providing more robust and context-aware imaging solutions.
9. The imaging device according to claim 1 , wherein the preset state includes a state in which a new object, which has not been present in a previous frame, appears and occupies a certain portion or greater of an image.
This invention relates to imaging devices designed to detect and analyze new objects appearing in a sequence of image frames. The problem addressed is the need for imaging systems to efficiently identify and process the appearance of new objects that were not present in prior frames, particularly when such objects occupy a significant portion of the image. The invention enhances an imaging device by incorporating a preset state detection mechanism that triggers when a new object appears in a frame and occupies a predefined area or larger. This mechanism ensures that the imaging device can promptly recognize and respond to the introduction of new objects, improving object detection accuracy and system responsiveness. The preset state may involve various conditions, including the object's size, position, or other attributes, to ensure reliable detection. The imaging device may further include processing components to analyze the detected object, such as tracking its movement, classifying its type, or triggering additional actions based on its presence. This solution is particularly useful in applications like surveillance, autonomous navigation, and industrial automation, where timely detection of new objects is critical. The invention ensures that the imaging device can dynamically adapt to changing environments by focusing on newly appearing objects, thereby enhancing overall system performance.
10. An imaging device for imaging an object, the imaging device comprising: an image pickup device configured to output an image signal, the image signal comprising a plurality of frames; a video signal processor configured to perform processing for generating a video signal based on the image signal; and an image recognition unit, implemented by hardware circuit, configured to perform a plurality of pieces of recognition processing on the image signal, wherein the processing of the video signal processor and the processing of the image recognition unit are executed in parallel to each other, and wherein a frame rate of the image pickup device is increased when the image recognition unit detects a preset state.
This invention relates to an imaging device designed to capture and process images of an object while performing parallel operations for video signal generation and image recognition. The device includes an image pickup component that outputs an image signal composed of multiple frames. A video signal processor generates a video signal from the image signal, while a dedicated hardware-based image recognition unit performs multiple recognition tasks on the same image signal. These two processing operations occur simultaneously to enhance efficiency. Additionally, the device dynamically adjusts the frame rate of the image pickup component when the image recognition unit detects a predefined state, allowing for adaptive imaging performance. This approach ensures real-time processing capabilities while optimizing resource usage based on detected conditions. The parallel execution of video processing and recognition tasks improves overall system responsiveness, particularly in applications requiring both high-quality video output and real-time object detection or analysis. The adaptive frame rate adjustment further enhances performance by increasing the capture rate when specific conditions are met, such as detecting motion or other significant events. This design is particularly useful in surveillance, automotive, or industrial imaging systems where both video recording and real-time analysis are critical.
11. An imaging device for imaging an object, the imaging device comprising: an image pickup device configured to output an image signal, the image signal comprising a plurality of frames; a video signal processor configured to perform processing for generating a video signal based on the image signal; and an image recognition unit, implemented by hardware circuit, configured to perform a plurality of pieces of recognition processing on the image signal, wherein the processing of the video signal processor and the processing of the image recognition unit are executed in parallel to each other, and wherein, when the image recognition unit detects a preset state, the image pickup device thins pixels and outputs only signals of specific pixels.
This invention relates to an imaging device designed to capture and process images while optimizing computational efficiency. The device addresses the challenge of balancing high-resolution imaging with real-time processing demands, particularly in applications requiring simultaneous video output and image recognition tasks. The imaging device includes an image pickup unit that generates an image signal composed of multiple frames. A video signal processor converts this signal into a video output, while a dedicated hardware-based image recognition unit performs multiple recognition processes in parallel. When the recognition unit detects a predefined condition, the image pickup unit selectively thins out pixels, outputting only signals from specific pixels to reduce processing load while maintaining critical recognition functionality. This parallel processing architecture and adaptive pixel thinning enhance performance by offloading recognition tasks from the main processing pipeline, ensuring efficient resource utilization without compromising video quality or recognition accuracy. The system is particularly useful in applications requiring real-time analysis, such as surveillance, autonomous navigation, or augmented reality, where both high-quality video and rapid recognition are essential.
Unknown
December 29, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.